Over the last decade, serious games became ‘serious’ educational tools; the idea of using the great strength of modern computer games for educational purposes experienced a significant boost. One, if not the main strength of computer games – used in serious domains – is seen in their motivational potential. Players play for fun and curiosity, for exciting and competition; the underlying serious purpose may be hidden or may be forgotten quickly. The challenge is that motivation is a highly individual construct; what is motivational to one player may bore another one; motivation depends on various psychological characteristics, stable traits (personality, cognitive abilities, preferences) and oscillating states (motivation, emotional state, moods). So, quite naturally, different players approach the games differently. Good games make an effort to support different skill levels for their players, in an attempt to provide an optimal experience. This adaptation can be extended to cover also different playing styles and even different player progression speeds. When dealing with serious games, especially educational games, the potential becomes even bigger, as it is also desirable to provide adaptation to different levels of prior knowledge and learning styles. All adaptive systems share the need to acquire knowledge about the user. The more the system knows a user, or the specific features of the user that are relevant to the system, the more accurate adaptation could be provided.
“Balancing” computer games according to certain (presumed) preferences and needs of the players in an autonomous and adaptive manner is an important feature. To be enjoyable, the game must match an individual player’s playing preferences, playing styles, and playing capabilities in a suitable way in order to avoid boredom or frustration be too difficult or too one-sided gameplay. So far there is almost exclusively a tradition of adaptively balancing recreational games. Main goal is to avoid undesired player emotions such as frustration (because the game is too hard) or boredom (because the game is too easy). Specifically in serious games, an appropriate adaptation is of crucial importance in order to reach and maintain fun and enjoyment on the one hand and effective, successful learning on the other hand.
Naturally, balance a conceptually complex environment such as a computer game (including visual, environmental, gameplay, game mechanics, narrative, and educational features) , requires a robust, psychologically meaningful yet simple assessment that can be realized in real time in the background of gaming and in addition to the computational demands of the game itself. Assessment, therefore, must be based on simple identifiable indicators and it must be based on valid heuristics; the indicators, thereby, can be divided into performance related aspects, emotional-motivational as well as personality related aspects. The performance related aspects include measuring/gathering, and analyzing/interpreting:
- Scores;
- Task completion rates;
- Task completion times;
- Task success rates (e.g., number of individual strikes in a shooting game);
- Task success depth (the quality or degree to which a task has been accomplished);
- Distances covered / progress in the game world;
- Exhibited knowledge, competence states, or skills;
- …
In addition to mere performance, it is important to account for oscillating psychological states (i.e., emotional states or, maybe more importantly, motivational states) and rather static psychological characteristics (i.e., personality traits). Assessment regarding emotional-motivational aspects include:
- Virtual character analyses (e.g., posture, alignment, locations, motions, etc.);
- Force exerted on interactive game controllers (if available);
- Facial expressions recorded by camera devices (e.g., webcams; if available);
- Gestures recorded by camera devices (e.g., Microsoft’s Kinect system; if available);
- Speech analyses – natural language processing;
- Text analyses;
In addition, recent developments account for the rich possibilities to rely on various psycho-physiological factors – even if such features and approaches are still in their infancy: Aspects considered by ongoing research are, for example, heart rate, respiration rate, or blood pressure. Finally, rather stable personality characters offer assessment indicators such as the adaptation and application of well-acknowledge personality inventories, for example, focusing on the so-called “Big 5” personality factors:
- Openness (inventive/curious vs. consistent/cautious)
- Conscientiousness (efficient/organized vs. easy-going/careless)
- Extraversion (outgoing/energetic vs. solitary/reserved)
- Agreeableness (friendly/compassionate vs. cold/unkind)
- Neuroticism (sensitive/nervous vs. secure/confident)
Adaptation and Balancing
There is a broad range of dimensions and techniques to respond or intervene in order to balance a game and tailor its manifestation to the individual learners. A very common technique is the so-called daynamic difficulty adjustment. This technique is widely used in entertainment games but also increasingly enters the AI of serious, in particular educational games. In simple terms, the idea is to increase the difficulty of the game or of game elements along with the increasing capabilities of the players. A similar approach is the so-called speed adjustment which actually focuses on racing-like games, where the speed of artificial opponents is adjusted with the player’s speed and racing abilities. In terms of learning, the exist ground-breaking work to adjust the difficult of learning materials (in mutual dependence with the global game difficulty and game characteristics). Also related to this type of balancing is the so-called rubber banding, which means artificially boosting performance (of a race car, for example) when falling behind.
A more sophisticated method, particularly when it comes to a learning-related adaptation and balancing, is problem solving support. This method attempts to identify where in a problem solving process (which is characteristic for many in-game tasks and quests) a player or learner is, to interpret whether support or guidance is required, and which type of support or guidance is the most appropriate one in the given situation. Techniques include hinting, feedback, cheer, guidance, giving examples, etc.
Examples of those balancing and adaptation techniques a sheer infinite. In general, each computer games has some elements of adaptation. Noteworthy are the results of two European projects. The first project was ELEKTRA, which focused on assessment and adaptation on an unobtrusive level and on subtle educational guidance. The second example, we want to highlight is 80Days. This project focused on advancing the non-invasive assessment and on appropriate psycho-pedagogical support of learning. The distinct novelty of the project was the integration of adaptive, interactive storytelling and adaptive game balancing.
How important is game balancing?
We have tried to outline the basic idea of game balancing and illustrate some examples of approaches to an intelligent automatic adaptation to the players and learners in terms of playing and in terms of learning. Of course this summary is not meant to be complete. It intention is primarily to raise awareness that mere mixing of learning materials with computer games and game-like features is likely not successful and may lead to, what famous Seymour Papert called a Shavian reversal, a chimera that unifies the worst from two worlds.
A smart, even intelligent individual balance and an emphasis on an autonomous, system-driven acknowledgement of individual needs and preferences, both in a game related as well as learning related sense is a key factor of a game’s success. This, again, concerns both the game hemisphere and the educational hemisphere.
To substantiate this claim, we conducted a meta-review in the field of game-based learning literature. We particularly looked into scientific articles presented results on the educational efficacy, the learning performance, of computer games. In total we analyzed over 300 papers published between 2009 and 2011 in SCI listed journals. On impressing result was that only a small percentage of those papers (i.e., about 10%) reported serious, non-trivial educational benefits from playing the games. Even more impressing was the fact that from those studies reporting reasonable and statistically significant benefits, 90% share that the investigated games bear some form of educational adaptation or personalization. This results (the details and further aspects will be published in the near future). This approach is in line with experimental findings that explicitly demonstrated that adaptation, personalization and subtle balancing results in superior gaming experience and educational gains.
In conclusion: To play is to learn and to teach is to balance. Good educational games inevitably need such intelligence personalization, adaptation, and balance.